An independent project ·Why this exists ·How this is made
Asheville · Western North Carolina
Step Up AVL

Step Up AVL · options paper

The Bus Trilemma

Three things are in play: how many places the bus system reaches, how many hours it runs, and how much the city pays for it. Asheville can keep two of the three steady, but not all three. Figuring out which one should change starts with two steps, and both of them are free.

Written 23 August 2026 Updated 25 September 2026 Figures from the National Transit Database, the city’s budget books, its 24 March and 28 July council documents and the signed operations contract, with every other source named at the foot of the page The working basis, margin and limit for each one The buses themselves The small print on the buses

Is there a solution?

Yes, but it takes several steps, not one fix. The first two steps cost nothing.

No option on the table lets Asheville keep all three of these at once: today's coverage, today's service hours, and today's local budget. One of the three has to change. Which one should change is a question the city can answer before it spends any money, and so far nobody has put that question in front of the council.

When the evidence points to a clear answer, this page says so. When it does not, this page says that too. Every number here has a basis, a margin, and a limit. All three are listed on the working detail, the companion page built for checking facts, not for reading straight through.

01Get the diagnosis right first

Asheville Rides Transit, or ART, costs less per rider than its peer systems. But it carries fewer riders per hour than it used to. A cost problem and a productivity problem need opposite fixes.

Two federal terms come up again and again on this page, so here they are first. A revenue hour is one hour a bus spends out on its route, ready to pick up riders. The federal count includes the short wait at the end of the line. It leaves out four things. One is deadhead, the driving a bus does outside revenue service, to and from the garage and between routes. The other three are operator training, maintenance testing, and any other use of the bus that is not carrying riders. A passenger trip is one boarding. If a rider changes buses once during a trip, that rider is counted twice. Nearly everything on this page is built from these two numbers, and from the money that pays for them. Where this page says riders per revenue hour, it means boardings per revenue hour. The federal count has no way to tell one rider making two boardings from two riders making one each. Asheville's fiscal year runs from July to June, so fiscal year 2026 ended in June 2026.

FY2024 · against 145 US bus systems in urbanized areas of 150,000–500,000
 AshevillePeer medianReading
Cost per rider$7.92$10.10below median
Cost per revenue hour$123.28$130.86below median
Cost per passenger mile$2.16$2.26below median
Cost per revenue vehicle mile$9.96$9.366% above
Riders per revenue hour15.5613.09above median

Four of the five measures put Asheville in a good spot. Three of the cost measures come in below the peer median, and riders per hour comes in above it. The fifth measure is cost per revenue vehicle mile. On that one, Asheville runs 6% above the median. So the federal data does not say that Asheville's buses cost too much. Here is what the data does say.

Asheville Rides Transit, fixed-route bus
 FY2019FY2026Change
Revenue hours81,02399,753+23.1%
Passenger trips1,978,7201,458,089−26.3%
Trips per revenue hour24.4214.62−40.1%

Revenue hours rose from 81,023 in fiscal year 2019 to 99,753 in fiscal year 2026. Passenger trips moved the other way, falling from 1,978,720 to 1,458,089. That means 23% more bus service is carrying 26% fewer boardings. The measure that combines both numbers falls the hardest: trips per revenue hour dropped from 24.42 to 14.62.

If a system costs too much, the fix is to buy less service. If a system is not carrying enough people, the fix is to make each hour of service carry more riders. Asheville's numbers point to the second problem. That means the cuts that would fix the first problem would actually make the second one worse. Why fiscal year 2024 is a local high point, and the two limits of this peer group →

02The math behind every option

Four things anyone can change, and one limit nobody has said out loud

total cost = (revenue hours) × (cost per revenue hour)
cost per rider = (cost per revenue hour) ÷ (riders per revenue hour)

Look at the two lines above. Every proposal anyone has made changes one of four things: how many hours the city buys, what each hour costs, how many riders each hour carries, and who pays the bill. The first three things are the terms in those two lines. The fourth, who pays, is not part of the math at all.

The fare riders pay is not a fifth term in this math. In fiscal year 2024, fares covered 4.7% of what the buses cost to run. Spread across every single boarding, the fare actually collected averages 37 cents.

All four of these are operating terms. That means they cover the cost of running the buses the city already owns. Buying new buses comes out of a different budget, called the capital plan. That is the part of the budget set aside for big purchases. See what the capital plan sets aside for buses →

The constraint

The 24 March 2026 council packet lists the cost of the operating contract's first four years. Year one costs $12,671,301. Year two costs $13,309,046. Year three costs $14,102,049. Year four costs $14,817,546. Each year costs more than the one before it. This page calls that yearly increase escalation.

The signed contract, released to us on 18 September 2026, gives the same four yearly totals. It fixes a monthly fee and a rate per revenue hour for each year, calls them firm fixed prices, and does not raise them automatically with an index. Several of its clauses can still move them, among them sustained inflation above what the contractor assumed and a change of more than 10 percent in the hours run. So the 5.35% a year figure used below is still the compound growth rate calculated from those four yearly totals, not quoted from the contract.

The contract is not the whole bill. The city also has $3,414,759 in costs outside the contract. Hold that amount flat, and total operating cost still grows 4.26% a year.

That 4.26% is the constraint. If cost per rider is to stay where it is today, ridership has to grow by 4.26% a year just to keep up. Over the three years from contract year 1 to contract year 4, that adds up to 13.3% more riders, none of which would count as real improvement.

03The trilemma

Coverage, service hours, local budget: pick any two to hold steady. The third one has to move.

Each row below holds two of the three steady and lets the third one move. A lit-up chip marks something that holds steady. A dim chip marks the thing that gives way. Every proposal anyone has made matches one of these three rows, and the council has only been shown one of them.

Coverage Service hours Local budget

Hold coverage and hours steady, and the local budget has to rise. Cost climbs 4.26% a year no matter what the council decides about individual stops. This is the conversation about new revenue, covered in Options G and H below.

Coverage Service hours Local budget

Hold coverage and the local budget steady, and service hours have to fall. Buses would still reach the same map of places, but there would be fewer of them, so waits get longer everywhere. Nobody has proposed this option. It is Option D below, and it saves between $1.34 million and $3.02 million a year: the low end at the contract's first-year prices, the high end only if every cost, the contract's fixed monthly fee included, falls with the hours.

Coverage Service hours Local budget

Hold service hours and the local budget steady, and coverage has to fall. The same total hours would be spread over fewer places. This is the route redesign the city's consultant drew up, and it is the only one of the three options the council has been shown.

The one thing that moves the whole triangle

One term does not trade against the others the way the rest do. Push cost per revenue hour down, and the whole triangle shifts, because every hour the city buys gets cheaper.

Asheville has already pushed this cost down. Its cost per revenue hour is $123.28, below the peer median of $130.86. Among the 47 peer systems that hire a private company to run their buses, Asheville is the 16th cheapest, against a median of $140.98 for that group.

So there is no easy saving left in the price of an hour, and the base contract sets that price through fiscal year 2030, subject to the adjustments it allows. The four option years that come after that are still open, and nobody has decided those yet.

04Every option, priced

The budget is $16,086,060. The rides are 1,458,089. Most of the options below just divide those two numbers a different way.

Cost per rider here is the FY2027 budget divided by FY2026's actual rides. The two figures come from different years, and the label says so every time it appears.
OptionCostPer riderEvidence it works
A. Status quo$16,086,060$11.03Grows 4.26% a year through year 4, and about $20.9m by the final option year at the contract's own option prices
B. Redesign delivers the consultant's +18%same$9.35No comparison city achieved this cost-neutrally
B2. Redesign delivers +22.4%, the size of Knoxville's FY2024 reboundsame$9.01Knoxville's first year after its August 2024 launch was −4.1%
C. Redesign delivers what Wilmington and Knoxville deliveredsame$11.50Wilmington −3.2%, Knoxville −4.1%
D. Shrink service back to FY2019 hourssaves $1.34m–$3.02m$8.96–$10.11Untested. Nobody has proposed it
E. Fare-free+$601,698n/aFY2024 fares; 3.7% of the FY2027 budget. Not a cost measure either way
F. Vans for the two smallest add-backs$53,270–$168,393n/a0.33%–1.05% of the budget
G. Quarter-cent county sales taxnew revenuen/aNeeds a referendum. Proceeds split per capita, and the county's own share counts only the people living outside every incorporated town and city in it, and only if the county itself runs or contracts transit. Spendable only on transit. Must supplement existing transit funding rather than replace it. Lost in New Hanover
H. Dedicated city property-tax slicenew revenuen/aWinston-Salem raised its slice 4.86¢ to 7.36¢; budgeted transit tax revenue rose $7,007,910, most of it from the rise, no referendum

The $16,086,060 budget is not a number the city can simply repeat next year. It only balances because the city is pulling $1,288,715 out of its fund balance, which equals 8.0% of the transit fund. Fund balance means money left over from earlier years, so spending it is, by definition, a one-time move.

The city's own budget book uses the word one-time, and it explains the draw as a result of expenses rising and revenue falling. That draw is about twice the size of the first year's escalation cost, which is $637,745. So there are really two separate asks here, not one. The first is replacing the $1,288,715. The second is paying for escalation, and only the second one is priced in the table above.

Option D is the outlier in the math, and nobody has put it on the table

Some of what a bus system spends goes up and down with the hours it runs. Driver pay, fuel, and the maintenance a bus needs from being driven all rise and fall with service. The rest does not change with service hours. That includes general administration, and the upkeep of buildings, stops, and equipment. So cutting hours never saves the full share of the budget that those hours seem to represent.

The city added 18,730 revenue hours since fiscal year 2019. Cut them back out, and the saving would be $3,020,379 a year, but only if every dollar of cost falls with the hours. It does not. The signed operations contract, released on 18 September 2026, splits what the city pays RATP Dev, the company that runs ART, into two parts. In the first year, one is a fixed fee of $510,019.64 a month and the other is $63.60 for each revenue hour run. The city buys the buses' fuel itself, about $8 a revenue hour in fiscal year 2024. So at the contract's first-year prices each hour cut saves about $71.68, and cutting those 18,730 hours saves $1,342,572. The contract's first-year price works out to about 103,000 revenue hours, more than the 99,753 run in fiscal year 2026, so going all the way back to fiscal year 2019's hours from the hours its prices imply would mean cutting about 22,000. A cut this deep may not keep those prices. One clause lets RATP Dev hold off on any service change of more than 10 percent in revenue hours until the two sides agree to raise the fee and the rate, and this cut is 18.8 percent. Another lets the city reduce service on 30 days' notice, and the contract does not say which of the two applies. Both numbers appear in the table below, because the gap between them is the whole point:

If shrinking back costs…TripsAll costs cutContract rate and fuel
0% of riders1,458,089$8.96$10.11
10%1,312,280$9.96$11.24
18.8%, proportional to the hours1,184,313$11.03$12.45
26.3%, the share ridership fell since FY20191,074,466$12.16$13.72

A break-even point here means the share of riders the city could lose and still end up at or below a certain cost per rider. There are two break-even points, and they sit far apart because they rest on different assumptions about which costs actually go down.

Assume every dollar falls with the hours. Then shrinking service beats today's $11.03 cost per rider as long as fewer than 18.8% of riders leave. It beats the redesign's best case of $9.35 as long as fewer than 4.2% of riders leave.

Now assume only the contract's hourly payment and the fuel fall, at the contract's first-year prices. Then the first threshold drops to 8.3%, and the second threshold disappears entirely. Even with no rider loss at all, shrinking still lands at $10.11 per rider, which is above the redesign's $9.35. See the contract's split and both break-even points →

Two cautions worth remembering

The 23% increase in service did not cause the 26% drop in riders. The pandemic caused that drop. And nothing here proves that the hours added since fiscal year 2019 are carrying nobody. What the data shows is that, across the system as a whole, the extra hours have not brought extra riders along with them.

Shrinking service is really the coverage debate again, just in a form that hides its true cost. Cut hours without cutting any destinations, and every route simply becomes less frequent. Cut destinations instead, and that is the redesign. There is no version of Option D where nobody loses something.

Option D is also the cheapest experiment available. The city could remove the least productive hours in one small pilot, publish trips per revenue hour every month, and let six months of numbers show whether those hours were carrying any riders at all.

Option F is cheap enough to break the deadlock completely

The council deadlocked over seven destinations. Together, those seven destinations carry 132 riders a day. That is out of the roughly 4,300 riders ART carries on an average day, so it comes to 3.1% of the system.

The city's 28 July 2026 council presentation offers four of the seven destinations back for a vote. The Outlets, Brevard Road and Transformation Village segment carries 40 riders a day. West Haywood, between Brevard and Patton, carries 34. MAHEC, a safety-net clinic, carries 11. The Social Security office carries 8. Three destinations are not on the add-back list: the Grove Park Inn at 30 riders a day, Kenilworth at 7, and Beaverdam at 2.

The two smallest of the four destinations on offer are the Social Security office, at 8 riders a day, and MAHEC, at 11 riders a day. Together that is 19 riders. The city could buy service to both of them at any of these prices:

HowA yearShare of the budget
Wilson-style shared microtransit, $11.17 a ride$53,2700.33%
Mountain Mobility, net of what it already recovers$127,4750.79%
Mountain Mobility at its gross cost per trip$168,3931.05%

Buying service to the two stops that stalled the decision costs between $53,270 and $168,393 a year. The $16,086,060 budget gets spent either way. What the council could not agree on was how to arrange that money, not how much of it to spend.

Buying the two stops out does not solve the cost problem, and nothing on this page solves that problem by itself. What it does is clear the way for a vote. The other two stops on the add-back list are a different case. The Outlets, Brevard Road and Transformation Village segment carries 40 riders a day, and under every assumption tested here, it is cheaper to serve with a bus than with a van. West Haywood carries 34 riders a day, and this page does not price out that option.

05The sequence

These steps run from cheapest to most expensive. The biggest number comes last because it is a decision the council still needs to make, not a purchase it has to fund.

STEP 0

Set a target for the money

A budget increase for a system with no stated goal is just an accounting decision, not a transit decision, because nothing afterward tells anyone whether the extra money actually worked.

ART has no published ridership target. It has no productivity target and no cost target either. The new contract does raise one standard: the monthly on-time-performance goal, from 72% to 80%. On-time performance measures reliability. It says nothing about how many people ride the bus.

The measure worth setting a target on already exists: trips per revenue hour. It is a federal measure and it is free to track. The federal database publishes the two numbers it is built from, rides and revenue hours, every month. Set a target, and the first check on progress arrives in about 90 days.

$0A council resolution and a monthly page on the city website

STEP 1

Request the two documents that would settle the coverage fight

Two documents would settle most of the coverage argument, and both are free to ask for.

The first is ART's Title VI Program. That is a compliance document rather than a service plan, and the service standards are one part of it among several. FTA requires direct and primary recipients to submit one every three years, and Asheville is one of them. The city is the designated recipient for the Asheville urbanized area: FTA sends the money here, and the city passes a share of it on to other local bodies. Its own list of those bodies is open and names seven, among them Buncombe, Haywood and Henderson counties. Passing money on is what makes a recipient a primary one, and a body can hold more than one of these roles at once. What decides the paperwork is that the city submits its own program to FTA, and the bodies it funds submit theirs to the city, on a schedule the city sets. The city says the document exists. Its transit pages tell readers to “see Asheville's entire Title VI program,” and they do not link it. Two parts of it are posted, in English, Spanish and Russian: a policy statement that sets out how to file a discrimination complaint, and the complaint form itself. The service standards are not. Ask for the rest, but expect it to answer less than the argument actually needs.

FTA Circular 4702.1B puts a larger set of duties on a provider that meets either of two conditions. The first is operating 50 or more fixed route vehicles in peak service while sitting in an urbanized area of 200,000 people or more, and both halves of that one have to hold. The second is being placed in the same category at the discretion of FTA's Director of Civil Rights, in consultation with the FTA Administrator.

Asheville's urbanized area clears the population half, at 285,776 people. ART's 19 vehicles in peak service do not clear the other half, and nothing suggests FTA has used its discretion here. So the larger set does not reach ART, and that set is not only the rider survey. It is also the demographic maps and charts, the monitoring of service, the three written policies a provider above the threshold has to adopt, on major service change, disparate impact and disproportionate burden, the first two of which its board has to approve as well, and the equity analysis it runs before a major service change or any fare change.

What does reach ART is every general requirement the circular puts on all recipients, and on top of that the duty to set and submit system-wide service standards and policies. One of the four required standards is service availability, which the circular describes as a general measure of the distribution of routes within a provider's own service area. That is the nearest thing to a written answer on coverage that ART has to produce.

And one duty reaches ART whatever its size. The circular says providers below the threshold are still responsible for the federal rules against disparate impact, and should review their policies and practices so that service and fare changes do not fall harder on people by race, color or national origin. A network redesign is exactly that kind of change. What ART is not required to hold is a written disparate impact policy, because that is one of the threshold-only items.

Because ART does not clear the threshold, it is not required to run the demographic rider survey the circular asks providers above it to repeat at least every five years. The survey taken for the network study drew 1,359 participants, riders and non-riders together, and they opted into it themselves rather than being randomly selected. The city's 2018 Transit Master Plan reports results from three earlier surveys of riders, from 2009, 2013 and the plan's own.

The second document is the route-level productivity data. The city has published it once, on page 50 of the Choices Report, as a scatter plot that labels each route but prints no values. The report's prose names one route, WE1, at almost 50 boardings per service hour. The rest can be measured off the picture to about half a boarding an hour, and a measurement taken off a chart is not a record. Boardings and revenue hours by route, as the city holds them, is exactly the input Option D needs, and exactly the input the redesign argument is missing.

That request asks for less than the city has. On 27 August 2024 the council approved buying automated passenger counters from a single supplier, Urban Transportation Associates, without taking other bids. Staff gave two reasons: the counters already on the buses are that company's, and it will not warrant another company's hardware; and staff were not sure it could program or support another company's counters. The city's own description of the devices is that they “count the number of passengers boarding and exiting at each stop.” Twelve of the thirty-five buses were running counters that could only send data over Wi-Fi; the replacements can use the cell network too, and the resolution says the data is then stored off site and “immediately available to staff for evaluation.” Boardings by route is that stop-level count added up.

$0Two records requests

STEP 5

Decide on the contract's option years now, not in 2030

The base contract totals $54,899,942 through year 4. Four more option years, worth $65,417,313, are still undecided. Adding those years brings the all-in total to $120,317,255, and pushes the final year's cost to $17,440,670. The option years come as two renewals of two years each, so the first decision covers years 5 and 6. If the city declines that first renewal, it returns to procurement in year 5 instead of year 9.

Even the base years are less locked in than they look. When asked about the contract on 24 March 2026, the city attorney told the council that the city always writes in a nonappropriation clause, and that his staff negotiates a termination provision as far as it can. The signed contract has both. Under the first, if the City Council does not fund the contract for any budget year, the city may end it without penalty on at least 30 days' written notice documenting the lack of funding. The city may also end the contract for its own convenience, in whole or in part, on 45 days' written notice; the contractor's claims are then settled under federal contract-termination rules.

The option years are the biggest number on this page, and it is not on the record who gets to decide them. The resolution authorizes the city manager to sign whatever the contract needs, and nothing in the packet says a renewal has to come back to the council. Asking who makes that call costs nothing, and it is the first question worth asking.

$65,417,313The decision window is open now

06What will not solve this

Four ideas that will get proposed anyway

  • Replacing the bus network with vans. Vans fall under a federal category called demand response. Looking at 126 peer agencies that run both fixed-route buses and demand response, the van service costs more per trip than that same agency's own bus service in every single one of them. Los Angeles Metro's own board report puts its Metro Micro van service at $42 a ride in September 2023, against a target of $20 to $25. It also records operational changes that cut the cost to $29.06 by the second quarter of fiscal year 2024. Sacramento cancelled its SmaRT Ride van service after the cost passed $47 a passenger, and narrowed the replacement service to seniors, disabled riders, and low-income riders to control the cost. Wilson, North Carolina, has the best general-public van cost in the state at $11.17 a ride, and that is still 41% more per rider than Asheville's buses already cost. Wilson also reports its data as a rural system, which puts it outside the peer group used everywhere else on this page. The closest comparison is in Asheville's own county: Buncombe County's Mountain Mobility costs $35.31 a trip, against ART's $7.92, in the same year and under the same reporting standard.
  • Treating the redesign as a fix for the budget. The plan itself says otherwise. The Draft Network Report tells readers, “this is cost-neutral, so if you want more service on your street, that must be paid for by cutting something else,” and it labels the 60%-larger Growth Network “not a proposal.” The numbers point the same way. At the consultant's own estimate of 18% more riders, the redesign cuts cost per rider by 15.3%. But escalation wipes out that saving in just 4.0 years. The better reason to adopt the redesign is reliability and job access.
  • Treating fare-free service as a way to cut costs. Fares collected in fiscal year 2024 totaled $601,698. That is 4.7% of that year's cost, or 3.7% of the fiscal year 2027 budget. Both numbers are true, and they are two different ratios measuring two different things. Going fare-free is a decision about access, not a decision about cost, whichever direction you look at it from.
  • Calling it an efficiency problem without asking efficient at what. Cost per revenue vehicle mile is the measuring stick behind that argument, and it is the one measure, out of the five in the first table, where Asheville comes out worse than its peers. It comes out worse mostly because Asheville's buses are slow. They average 12.38 miles per hour, against Roanoke's 16.29 miles per hour. An hour of driving in Asheville simply covers fewer miles, so the cost of that hour gets spread across fewer of them. Measured by cost per rider instead, Roanoke's buses cost within one cent of Asheville's.

07Three things nobody has proven

And each one would decide the answer

  1. Whether the hours added since fiscal year 2019 are actually carrying any riders. Route-level boardings per revenue hour would answer this question, and the city already has that data. Its buses count boardings and exits stop by stop.
  2. What Asheville's riders actually need. No weighted survey of riders exists, and federal rules do not require ART to run one. Whatever the Title VI Program contains is the closest thing that exists, and it may say nothing about riders at all. Filling this gap probably means commissioning a new survey, not simply requesting an existing document.
  3. Whether a cost-neutral redesign can raise ridership at all. The two real-world cases checked here both went the other way. Knoxville lost 4.1% of its riders in the twelve months after its August 2024 launch, even though it added 22.8% more service. Wilmington lost 3.2% of its riders across the calendar years after a redesign that protected coverage in writing. Two cities cannot settle this question, and Asheville is neither of them. But two cities are the only evidence available, and it points toward adopting the redesign for its reliability and access benefits, while treating any ridership gain as a hope rather than a plan. See both Wilmington time periods, and what Wave's own staff claim →
One line worth remembering

Eighteen trips per revenue hour can be reached two different ways, and both reach the same target. Either carry 23% more riders on today's amount of service, or carry today's number of riders on 18.8% fewer hours. Today's 1,458,089 riders, divided by fiscal year 2019's 81,023 hours, comes out to just under 18 riders an hour (17.996).

But these two paths do not cost the same. Growing ridership to reach 18 lands at $8.96 a rider, and this does not depend on which costs are fixed. Shrinking hours to reach 18 also lands at $8.96, but only if every dollar of cost falls along with the hours. If only the contract's hourly payments and the fuel fall, it lands at $10.11 instead. That is at the contract's first-year prices, which a cut this deep may not keep, and whether anything else falls with the hours is not established. And Asheville has published no target on either side of 18.